CFD - Need Help With Mesh Set Up and Simulation Project

Project Support
1

Hello! Thanks so much for helping out.

I have a school project due Friday and I need to finish this simulation by today or tomorrow. Right now I’m having trouble setting up a CFD simulation. I followed the tutorial “Aerodynamics Simulation of Flow Around a Vehicle” and instead of 20 m/s, used 7 m/s (my design is for a glider, which is slower). However, because of the size difference, I did not follow the region refinement settings. I have been unable to create a usable mesh for a simulation run.

I tried refining the CAD model multiple times, adding a fillet to the edges of wings, but the mesh quality visualization (isovolume) still shows a high aspect ratio in the tail. In the screenshot you can use on the end that there is a slight gap, so the fillet is limited because it cannot cut into a new body. I filleted as much as possible, but it still gives high aspect ratio here.

Annoying
Annoying1016×761 52.3 KB

Annoying2
Annoying21023×773 40.8 KB

I have reviewed forums, tutorials, and help articles but am left confused with the overload of information. I made many different meshes trying to tweak settings (experimenting with inflating boundary layer, local element size, closer region refinements), but as yet, none have fit within acceptable parameters.

The best mesh (Mesh 18) I have created was ~.50 accurate, but the simulation (Run 4) broke down because it used too much memory and from what I could see from the solver log, the pressure was uniform. The goal of this project is to see the aerodynamic behavior of my glider and look at different pressure regions. I need to see which areas have the most drag, plus get some cool visuals.

I copy/pasted some of my data in this mesh log table, but recent meshes after Mesh 18 have not yet been compiled. The first two columns are maximum CFD and recommended CFD values.

Mesh Data
Mesh Data1130×463 86.5 KB

I need to lower AspectRatio and NonOrthogonality to improve mesh quality. Ideally a different workflow is available that does not include refining the CAD model (I have to use Fusion 360, then Onshape, then import and go through creating a new flow region).

Please tell me the exact settings I can use to make a usable mesh and what I need to change for my simulation to be effective. No need to be instructive, or guide me to an answer, just tell me what I need to do exactly!

Higher wind speed? Is my simulation flow region too large for my small design? Thank you for your help!

2

Hello!

Is it a requirement that you run this as a transient simulation?

image
image800×344 18.8 KB

Transient simulations tend to require quite a lot of extra caution during the setup, and even then can take much longer than steady-state runs. The tutorial that you are referring to contains a steady-state analysis, which would be my recommendation for an initial study.

Cheers

3

@RicardoParis

Thanks for the reply. Started using SimScale last week, so not entirely sure what the difference is. I will change the setting, though. Will this resolve either the issue with my mesh quality or the run taking too long (not enough memory)? My concern is that if I do not address mesh problems first, I will have one less run remaining (4/10 left with my account plan) and stuck.

Appreciate your help and any other suggestions.

Thanks again.

4

When it comes to stability: it’s difficult to say for sure without running a study, but it’s definitely going to be much more stable as a steady-state case.

Regarding runtime/memory: this really depends on the mesh size. You have 18 million cells right now, which is quite a lot. For a community license, if you can bring this number down to 6-10 million~ it might be a bit safer.

5

@RicardoParis

Got it! That makes sense, thanks for the clarification!

I have 20+ different meshes due to so much tweaking and experimenting as well as unfamiliarity with SimScale. The majority are under 18 million cells (actually less than 6 million, too). If I run a new time dependent simulation with these “small” meshes I am sure the velocity fields will diverge.

What can I do specifically to make a new mesh—get in the sweet spot of 6-10 million, lower the aspect ratio, improve nonOrthogonality, mesh quality, etc? Is there any chance you can give me specific settings for region refinements/local element size/inflate boundary layer?

I understand this might be difficult or time consuming on your part, but I’d really appreciate it. Thank you for your help so far!

6

To be honest I don’t think you would be able to run a time dependent (i.e. transient) run with a community account - this will be too heavy to run given the mesh size.

I do think that mesh 33 has a decent chance of running successfully for a steady-state simulation though. The tricky part to mesh tends to be the trailing edges if they are too pointy.

You will notice that an usual approach is to ‘cut’ the trailing edge a little bit short to prevent the sharp corner from arising. E.g. see the trailing edge in this geometry.

Cheers